Removing material from a substrate to form microscopic or nanoscopic structures is referred to as micromachining, milling, or etching. Lasers beams and charged particle beams are two particular technologies used for micromachining. Each has advantages and limitations in various applications.
Laser systems use several different mechanisms for micromachining. In some processes, the laser is used to supply heat to a substrate to induce a chemical reaction. The reaction occurs only in the heated areas. The heat tends to diffuse to an area larger than the laser beam spot, making the resolution of the process poorer than the laser spot size. Another mechanism used in laser micromachining is photochemical etching, in which the laser energy is absorbed by individual atoms or molecules of the substrate, exciting them into a state in which they can chemically react with an etchant. Photochemical etching is limited to materials that are photochemically active. Another mechanism used in laser machining is laser ablation, in which energy supplied rapidly to a small volume causes atoms to be expelled from the substrate without heating the substrate. Laser ablation using a fast-pulsed, femtosecond laser is described, for example, in U.S. Re. 37,585 to Mourou for “Method for Controlling Configuration of Laser Induced Breakdown and Ablation.” Femtosecond laser ablation overcomes some of the limitations of the processes described above.
Charged particle beams include ion beams and electron beams. Ions in a focused beam typically have sufficient momentum to micromachine by physically ejecting material from a surface. Because electrons are much lighter than ions, electron beams are typically limited to removing material by inducing a chemical reaction between an etchant vapor and the substrate. Ions beams typically are generated from a liquid metal ion source or by a plasma ion source. The spot size of a charged particle beam depends on many factors, including the type of particles and the current in the beam. A beam with low current can typically be focused to a smaller spot and therefore produce a smaller structure than a beam with high current, but a low current beam takes longer to micromachine a structure than a high current beam.
Lasers are typically capable of supplying energy to a substrate at a much higher rate than charged particle beams, and so lasers typically have much higher material removal rates than charged particle beams. The wavelength of lasers, however, is much larger than the wavelength of the charged particles in the charged particle beams. Because the size to which a beam can be focused is, in part, limited by the beam wavelength, the minimum spot size of a laser beam is typically larger than the minimum spot size of a charged particle beam. A. P. Joglekar et al, in “Optics at critical intensity: Applications to Nanomorphing,” Proceedings of the National Academy of Science, vol. 101, no. 16, pp. 5856-861 (2004) (“Joglekar et al.”) shows that features smaller than the wavelength can be achieved using laser pulses shorter than about 10 picoseconds near the critical intensity for ablation. The impact of small deviations in pulse energy on the resulting feature makes this process difficult to repeat consistently.
While a charged particle beam typically has greater resolution than a laser beam and can micromachine an extremely small structure, the beam current is limited and the micromachining operation can be unacceptably slow. Laser micromachining, on the other hand, is generally much faster, but the resolution is inherently limited by the longer beam wavelength.
One way to take advantage of both the faster micromachining capability of lasers and the higher precision of charged particle beams is to sequentially process a sample. Sequential processing is described, for example, by M. Paniccia et al. in “Novel Optical Probing and Micromachining Techniques for Silicon Debug of Flip Chip Packaged Microprocessors,” Microelectronic Engineering 46 (pp. 27-34 1999) (“Paniccia”). Paniccia et al. describes a known technique for accessing the active portion of a semiconductor flip chip using laser-induced chemical etching to remove the bulk of material, and then using a charged particle beam for the final, more precise micromachining. A problem with sequential processing is determining when to stop the faster, less precise laser micromachining and to begin the more precise charged particle beam processing. If the laser processing is stopped too soon, excess material will remain for removal by the charged particle beam. If the laser processing is stopped too late, the work piece will be damaged. Determining when to stop processing is referred to as “endpointing.” While there are several known methods for detecting when a micromachining process cuts through a first material to expose a second material, it is typical to stop laser processing before a change in material is reached, and so determining the end point is more difficult.
While sequential processing—laser processing a sample followed by particle beam processing—has its advantages, there remain problems of the endpointing and of the relatively low resolution of the original laser processing.